Encrypted login | home

Program Information

A New Intra-Fractional Prostate Motion Tracking Method in Volumetric Modulated Arc Therapy (VMAT) Via 2D/3D Registration

Y Chi

Y Chi1*, N Hassan Rezaeian1 , R Hannan1 , X Jia1 , (1) UT Southwestern Medical Center, Dallas, Texas


SU-G-JeP1-2 (Sunday, July 31, 2016) 4:00 PM - 4:30 PM Room: ePoster Theater

Purpose: Intra-fractional prostate motion leads uncertainty on delivered dose in radiotherapy and may cause significant dose deviation from the planned dose distribution. This is especially a concern in scenarios with a high dose per fraction and hence a long delivery time, e.g. stereotactic body radiotherapy. Knowledge about intra-fractional prostate motion is valuable to address this problem, e.g. by reconstructing delivered dose and performing adaptation. This study proposes a new approach to determine intra-fractional prostate motion in VMAT via 2D/3D maker registration.

Methods: At our institution, each patient has three markers implanted in the prostate. During treatment delivery, kV triggered images were taken every three seconds to acquire 2D projection of 3D anatomy at the direction orthogonal to the therapeutic beam. Projected marker locations were identified on each projection image using template matching with geometric constraints. 3D prostate translation and rotation for each triggered image were obtained by solving an optimization problem, such that the calculated marker locations match the measured ones. Inter-image motion smoothness was employed as a constraint. We tested this method in simulation studies with five realistic prostate motion trajectories acquired via Calypso and in real phantom experiments.

Results: For the simulation case, the motion range for these patients was 0.5~6.0 mm. Root mean square (RMS) error of calculated motion along left-right (LR), anterior-posterior (AP) and cranial-caudal (CC) directions were 0.26mm, 0.36mm, and 0.016mm, respectively. The motion range in the phantom study along LR, AP, and CC directions were 15mm, 20mm and 10mm. The mean RMS errors along these directions were 1.99mm, 1.37mm and 0.22mm.

Conclusion: A new prostate motion tracking algorithm based on kV triggered images has been developed and validated. Clinically acceptable accuracy has been achieved.

Contact Email: